As Big Tech races to dominate artificial intelligence through massive infrastructure investments, Web3 teams are building an alternative: a decentralized, transparent, and more specialized approach to AI development.
In 2024 alone, Amazon, Google, Meta, and Microsoft reportedly spent around $180 billion on AI-related infrastructure and data centers. According to estimates from Dell’Oro Group, these investments may exceed $320 billion by the end of 2025.
Despite these figures, Web3 teams believe they are positioned to carve out a unique space in the AI ecosystem. At EthCC’s Agents Day, we spoke with Clara Tsao (Co-founder, Filecoin Foundation), Dan Wang (CEO, Aethir), and Brandon Slake (Head of Protocol Partnerships, Messari) to understand how Web3 is rethinking AI.
Infrastructure Challenges: Cost, Complexity, and Scale
Building AI systems is not just expensive—it’s structurally complex. Clara Tsao notes that effective AI requires a fully integrated stack: data layers, compute resources, GPUs, and interoperable agents.
Key infrastructure pain points:
- Time-to-market and scalability: Most AI services face long lead times and difficulty scaling effectively.
- Cost of deployment: According to Dan Wang, even promising AI projects often fail to scale due to unsustainable operating costs and unclear monetization models.
Big Tech’s advantage is clear—they can absorb costs, optimize at scale, and have deep control over user ecosystems. Web3 lacks these resources but brings a different value proposition.
“Our main competition isn’t just other AI startups—it’s traditional cloud providers with massive infrastructure and market dominance,” says Tsao.
Still, momentum is building. The decentralized physical infrastructure networks (DePIN) sector—where Filecoin is a leading player—has already surpassed $40 billion in capitalization. Some forecasts predict this could grow to $3.5 trillion by 2028 (World Economic Forum).
Trust and Data Integrity
Cost isn’t the only obstacle. The quality and reliability of data are becoming critical, especially in the context of AI training and deployment.
Core concerns include:
- Data verifiability: Traditional cloud services often obscure the provenance and integrity of stored data.
- GIGO principle: “Garbage in, garbage out” is especially relevant for fast-scaling AI applications processing massive, unverified data volumes.
Web3 infrastructure introduces a compelling solution: cryptographic proofs and transparent storage. By leveraging blockchain, developers can validate both the source and integrity of the data.
“Verification is no longer optional. In a world of deepfakes and synthetic content, knowing where your data comes from is essential,” says Slake.
Web3’s native data architecture—through tools like Filecoin—enables:
- Decentralized storage
- Proof-of-origin
- Immutable audit trails
Strategic Specialization Over Generalization
Another strength of decentralized AI is the focus on specialization. While Big Tech aims to deliver general-purpose models, Web3 players are optimizing for niche applications.
“Think of it like medicine,” explains Slake. “General AI models are like general practitioners. But when precision is required—especially in fields like crypto or finance—specialist agents outperform.”
Emerging specialization examples:
- Messari Copilot: Designed specifically for filtering and analyzing crypto data.
- DeFi trading agents: Optimized for executing tasks in decentralized markets with minimal human oversight.
General-purpose LLMs like ChatGPT will remain relevant, but Web3 agents offer sharper, domain-specific performance in critical verticals.
Web3 + AI: Where the Growth Is
AI agents operating on-chain are growing rapidly. According to VanEck, over one million blockchain-based AI agents could be deployed by the end of 2025, especially in the financial sector.
“We’re seeing the most aggressive adoption of AI agents in DeFi, where automation is key,” notes Tsao.
Why data storage matters:
- AI systems—centralized or decentralized—are only as good as the data they train on.
- Long-term data integrity and accessibility are core enablers for scalable AI deployment.
“Data is foundational. Without reliable storage, even the best AI models collapse under real-world use,” Tsao concludes.
Final Thoughts: Can Web3 Shift the AI Balance?
The AI landscape is still in its formative stage. While Big Tech builds vertically integrated empires, Web3 offers a decentralized, auditable, and specialized alternative.
With growing investment in decentralized infrastructure, new data provenance standards, and use-case-specific agents, Web3 may not need to "beat" Big Tech outright—it just needs to outmaneuver it where it matters.
As history shows, the right combination of timing, focus, and adaptability can tip the balance—even when the odds seem one-sided.
Top comments (0)